白细胞图像的自动分割算法_顾广华.pdfVIP

第 卷 第 期 仪 器 仪 表 学 报 Vol.30 No. 9 30 9 2009 年 9 月 Chinese Journal of Scientific Instrument Sep. 2009 * 白细胞图像的自动分割算法 顾广华，崔 冬 （燕山大学信息科学与工程学院 秦皇岛 066004 ） 摘 要：根据白细胞核在其饱和度分量 S 和绿分量 G 中的分布特点，构造细胞变换图像，提取细胞核，实现白细胞的初步 定位。对存在细胞粘连的细胞图像，利用细胞边缘的相位角变化来检测细胞粘连区域并除之，采用线性插值法拟合边缘曲线， 得到封闭的白细胞边缘轮廓，最终提取白细胞。本算法可自动分割白细胞，具有很好的分割效果。 关键词：白细胞分割；变换图像；细胞粘连；相位检测；线性插值 中图分类号： 文献标识码： 国家标准学科分类代码： TP391.4 A 510.2040 Automatic segmentation algorithm for leukocyte images Gu Guanghua, Cui Dong (School of Information Science and Engineering, Yanshan University, Qinhuangdao 066004, China) Abstract ：The paper constructs a transpositional cell image based on the saturation and green component images according to different distribution characteristic of leukocyte nucleus in the two images. The leukocyte nucleus can be extracted from the transpositional image and the initial location of the leukocyte can be achieved. For the overlapped cell images, firstly the overlapping regions are discovered from the phase change of the leukocyte contour and eliminated. Then the edge curve is re-constructed to get a closed contour using linear interpolation. Finally, the leukocyte can be extracted successfully. The proposed algorithm can be used to process the leukocyte automatically and achieve good results of leukocyte segmentation. Key words ：leukocyte segmentation; transpositional image; cell overlapping; phase detection; linear interpolation 像。然而血细胞显微图像中经常会出现细胞重叠和粘连 1 引 言 的现象，如何分解粘连细胞是一个亟需解决的问题。